Zero to OEE in an Hour

Watch Webinar

Webinar Overview

Are you interested in optimizing your manufacturing operations and boosting productivity? Watch this webinar to discover how you can quickly and effectively implement, measure and improve Overall Equipment Efficiency (OEE) to get performance improvements by 10% to 30% within the four walls of your manufacturing set-up.

OEE is a vital metric for measuring the effectiveness of your manufacturing processes and equipment. By understanding and improving OEE, you can identify areas of improvement, reduce downtime, increase throughput, and ultimately enhance overall productivity.

During this webinar, experts from HiveMQ, 4IR, and Riveron guide you through the process of implementing, measuring and improving OEE, even if you’re starting from scratch. Here are a few key takeaways from this webinar recording:

• Get practical insights, best practices, and real-life examples that will empower you to transform your manufacturing operations.

• Learn how to build a Unified Namespace (UNS) that models and aggregates OEE data across all of your manufacturing operations.

• Watch a live demo of how to build an OEE dashboard in an hour, showcasing KPIs that your executives need for improving ROI and getting faster time-to-value.

This webinar is designed to equip plant managers, operations professionals, and digital transformation specialists with the knowledge and tools needed to kickstart their journey toward production-grade OEE. Our goal is to help streamline operations, reduce costs, and maximize manufacturing efficiency.

Key Takeaways

• Overall Equipment Effectiveness (OEE) is a key metric for optimizing production and can help manufacturers boost ROI, workforce productivity, and on-time delivery. It is calculated based on availability, performance, and quality.

• The webinar covers how to build a Unified Namespace (UNS) to model and aggregate OEE data across manufacturing operations, and how to build an OEE dashboard from scratch in an hour.

• World-class OEE is 85%. Typical manufacturing OEE is around 60%, with opportunities for quick improvement to 85% with Industry 4.0 approaches. Challenges to calculating OEE include human errors in data entry, connectivity/bandwidth constraints, cybersecurity threats, lack of scalability, and incomplete data analysis.

• HiveMQ provides a reliable real-time data backbone to move industrial data from the edge to the cloud at scale. Key features enabling this include clustering, data replication, persistence, and protection mechanisms.

• 4IR Solutions' FactoryStack managed platform makes it easy to deploy cloud environments for manufacturing use cases like OEE monitoring.

• Riveron provides end-to-end consulting including OEE dashboard development, connecting devices, integrating platforms, establishing governance for scaling pilots, and driving organizational change management.

• MQTT is gaining traction as a protocol for industrial use cases due to its lightweight, publish-subscribe, bi-directional messaging model. 

• OEE initiatives are often driven top-down as part of digital transformation efforts, coupled with pressure for manufacturing data integration.

• Many platforms exist for building OEE visualizations; when evaluating options, focus on aligning with organizational requirements and team capabilities.

Transcript

Introduction

Jayashree Hegde: 00:00:08.471 Hello, everyone. Good morning. Good afternoon. Good evening. I'm Jayashree Hegde, and it is my pleasure to welcome you all to this exciting joint webinar with 4IR Solutions and Riveron. Today, we have industry product experts who will guide you all on how to build a Unified Namespace that models and aggregates OEE data across your manufacturing operations. Not only that, but they will also have a live demonstration showing you all how to build an OEE dashboard from scratch. So make sure to stay tuned with us till the end of the session as you are going to gain a wealth of insight in just an hour. Allow me to quickly introduce you all to our speakers for today. We have James Burnand, the CEO of 4IR Solutions. Joining him is Joe Dolivo, the CTO of 4IR Solutions, and Remus Pop, the Technical Director of Riveron. And we have my dear colleagues, Ravi Subramanyan, Director of Industry Solutions Manufacturing at HiveMQ, and Ryan Dussiaume, Solutions Engineer at HiveMQ. I will let them introduce themselves later. Before we dive into this exciting session, let's go over a few housekeeping items. Firstly, we are recording this session, and we will share the recording in a follow-up email. Feel free to submit your questions in the Q&A box, and we will run a poll during the Q&A. I request you all to participate. Now, without further ado, I will hand it over to Joe. Welcome, everyone.

OEE Demo Provisioning in Real-Time

Joseph Dolivo: 00:01:42.416 Awesome. Thanks so much. And thanks, everybody, for joining. So in the interest of time, we're going to get started by actually provisioning this demo, since it's going to take some time to provision everything from scratch. And I'm going to walk through and show you what that looks like. And then we'll go through the rest of the introductions and the rest of the agenda today. So we are going to do something pretty novel in that we're going to provision this demo from scratch in less than an hour and be able to demonstrate it to you. So what does this look like? So we have two production lines. These are simulated, but these are based off of real lines that are generating things like part counts, machine statuses, rejects, things like that. Pretty standard data for a lot of different equipment. And these machines are communicating with the PLC. We have an actual Opto 22 groov EPIC device that is running a version of Ignition that's basically communicating with this device. And it's sampling all of the data points that are going to be used for the OEE calculations. And so this is what we have right now running. So you can imagine these machines are up right now.

Joseph Dolivo: 00:02:42.782 The part that we're going to be provisioning today during the live demo is all of the stuff inside of the cloud. So we'll talk a little bit more about what the details are inside of here. This is a product and managed service that we provide called FactoryStack, really built around HiveMQ as the centralized tool that enables the Unified Namespace. And we have a number of applications that we run inside of this stack, including Ignition. And what you'll see is that we've got HiveMQ here, which is communicating down with Ignition running inside of the groov EPIC over the MQTT protocol. And then Ignition is communicating with this HiveMQ instance also over MQTT. And then there's a database connection that's created and provisioned automatically when this is spun up. And then we have some support services that we'll get into more later, such as version control, Grafana for graphics and monitoring of dashboards and logging. And then also with the demo, once this is provisioned, we're actually going to launch it. My colleague Remus is going to launch this, and we're actually going to demonstrate live data coming in and what the dashboard looks like as a result of that. And you'll notice we've highlighted a couple of physical geographies here. So this physical device is located in Pennsylvania sitting next to James.

Joseph Dolivo: 00:03:57.237 The cloud infrastructure that we're going to provision dynamically is all going to be inside of US East. They've got a data center in Virginia. And then Remus is sitting over there in Michigan to launch this. So really just describes kind of the breadth and scale of what we're able to do with the cloud. So without further ado, I'm going to actually go ahead and start this demo so that we have enough time to have it provisioned. And I'll show you what that looks like. So if I get out of full screen here, and I'm going to toggle over to this application, which we call the Designer. So this is a tool that we use internally to help us construct custom FactoryStacks, as we call it, for each of our customers. And so, again, central to all of this is HiveMQ. So picture this like building blocks that I can drag-and-drop and configure. So for purposes of the demo, I'm going to go ahead and configure a HiveMQ cluster to have two nodes. And I'm going to give it two virtual CPUs and four gigs of RAM because we have a pretty small system that we're demonstrating here. I'm going to go ahead and save this.

Joseph Dolivo: 00:05:01.045 And then connected to this, and this is what I kind of was showing inside of the diagram, I'm going to have an instance of Ignition, which is going to connect down to the broker. And this one I'm going to configure with a version of Ignition. I'm going to upload a gateway backup. So this basically has some of the screens already built and pre-configured. So this is going to get uploaded and basically loaded when we instantiate all the cloud infrastructure. And then similarly, I'm going to select some CPU and RAM resources. And I have this very nice visual configurator that is allowing me to configure each of these tools and then to connect them, which is going to do some things in the background like setting up connections. And two more pieces I'm going to put on here. I'm going to drag in a database. And so the Ignition application that's going to be collecting and generating the OEE numbers is actually going to require a connection to a database. So similarly, I'm going to come in and give it a single node replica. I'm going to provide some CPU and RAM for this instance. And then lastly, I'm just going to have an instance of Git that's connected.

Introduction to Speakers: HiveMQ, 4IR Solutions and Riveron

Joseph Dolivo: 00:06:01.179 And this is basically going to allow me, out of the box, to version control my OT application that I've developed inside of Ignition, which is pretty novel still in this space. And that's pretty much all I'm going to have to do to get this up and going and get this provisioned. So it's pretty nice that once we kind of have this nice visual configurator, all I have to do is to click Deploy. And what you'll see is this is going to launch into our internal instance for basically running these deployment pipelines. And you can see here now that I've got this pipeline, has now been basically staged, and this is going to be running in the background as we go through the rest of the presentation. So I'm going to switch back to this slide, and then we'll just continue on here with introductions. So I'll start with myself because I've been talking. So I'm Joe Dolivo. I'm the CTO of 4IR Solutions. I've been working in the manufacturing industry for a long time. And now we're doing some interesting things with cloud and hybrid cloud. I'll pass it over to my colleague, James, who's our CEO, and we'll kind of go left to right and then top to bottom. So James, take it away.

James Burnand: 00:07:10.528 Should have been ready on the unmute button. Thanks, Joe. So I'm James Burnand. I'm the CEO of 4IR Solutions. I've been in the manufacturing industry just a little bit longer than Joe has. You can maybe tell by my hair being a little grayer. I come from a background of systems integration, and I did a lot of work in the cybersecurity space prior to working with 4IR for the last couple of years. So, Remus.

Remus Pop: 00:07:33.072 Hey, everybody. My name is Remus Pop. I am the Technical Director of the Intelligent Manufacturing Solutions Group at Riveron. We are a national business advisory firm. My background, also long in manufacturing, spent across a couple of different industries like automotive, battery manufacturing, aerospace. And yeah, just really looking forward to being a part of this and look forward to showing off what we can do.

Ravi Subramanyan: 00:07:55.005 Thank you, Remus. Let me introduce myself. This is Ravi Subramanyan. I work for HiveMQ. HiveMQ provides an enterprise-grade MQTT-based software broker, and we are happy to be hosting this webinar and doing this demo. So let me hand it over to Ryan.

Ryan Dussiaume: 00:08:13.626 Hey, everybody, Ryan Dussiaume. I'm a solution engineer here at HiveMQ. And yeah, my background is in things like hybrid cloud, SaaS, Kubernetes and cloud native software applications previously before joining HiveMQ, where I'm now working quite a bit with messaging platforms and the MQTT protocol. So looking forward to seeing this demo and how we can provision this in just an hour.

Ravi Subramanyan: 00:08:47.803 Yeah. All right, okay. Let me jump to this field on this. Okay, so quickly, agenda, right? So obviously the topic is OEE in an Hour. So Joe already started the environment, so the timer is on. So we've done the provisioning in real time. I'm going to be kind of talking about what is OEE? How is it calculated? Why is it important? What are some of the challenges to doing OEE? And then we're going to show some architecture diagram which Ryan is going to do. Show architecture around what we're going to show, and in general, how what we're going to show, for example, scales beyond the demo, for example, right? And then that's kind of where the different technologies come together, from Riveron, from 4IR, and HiveMQ working together in this Industry 4.0 architecture for being able to track and also being able to increase OEE. And we're going to be talking about that as well. And then an OEE application demo, and then finally we're going to do a Q&A. There'll be plenty of time for that, so please keep your questions coming.

What is OEE and How Do We Calculate It?

Ravi Subramanyan: 00:09:51.837 All right, so importance of OEE, right? So just for definition, OEE stands for Overall Equipment Effectiveness. And it is a key performance indicator in a factory environment, which typically faces issues with labor shortage and machine downtime. So OEE provides a quick way for manufacturers to be able to ship products on time and be able to quote prices despite all these issues that they have, right? So that's kind of at a high level. How is it calculated? There are multiple ways to do it. There's a simple way, which is basically using the goods count times the idle cycle time divided by the planned production time. That's kind of a simple way to do it. But the more common way and the more preferred way to do it is based on availability, performance, and quality. Basically adding these different parameters to get the preferred OEE. And so there's a tight correlation between OEE and effective production, right? So in an ideal world, I guess perfect production is 100% OEE where you're only producing high-quality components and you literally have no downtime.

Ravi Subramanyan: 00:11:01.550 Now, obviously, that's kind of out there. Everybody's striving for it. But really, the world-class for discrete manufacturing is kind of the real objective, right, attaining 85% OEE. But typically, what you see, the most common you see is like 60% OEE, right? There is kind of a lot of opportunities to improve. And this is kind of like OEEs that have started implementing some of the Industry 4.0. Really, the low score is considered poor, but it's not unusual for manufacturing because they've just started on this path of performance tracking and improvement. And very quickly, they can go from 40 to 60 to even 85, right? So again, some of the key benefits is ROI. So significant impact on customers' bottom line, ability to produce more products on the same equipment at the same time. From a workforce productivity perspective, it helps companies figure out when downtimes are occurring, right? So they get a view on the productivity statistics at different levels within the organization, and they can start highlighting some of the delays and the causes for that, right? "How much time does it take for setup and handovers and changeovers and things like that?"

Ravi Subramanyan: 00:12:20.684 And all these can be visualized, right? And Remus is kind of going to talk about that, right? I mean, you have the technology. How can you build the people and process behind ensuring that this leads to improving OEE? And some of the challenges, though, is there's still kind of human errors, right, because after all, as human beings collecting this information, because of work pressures, operators are overwhelmed and information can be incorrectly entered, right? Typical factory environment, you have connectivity and bandwidth challenges. So the important data that you need for being able to track and collect the OEE could be not available because of constrained environments. There's always cybersecurity concerns, right? There could be hacking that can happen due to which you may not be able to have the information that you need. Lack of scalability, right, where, yeah, you can implement OEE in a small scale, but as you kind of have more systems coming in, that's kind of where you start seeing the issue in terms of the number of connections. Incomplete analysis is like sampling the data and then basically saying, "Hey, I don't have access to all of my data, but I'm going to sample a portion of it. And then I'm going to predict how things will work." That doesn't quite work because factory systems are so diverse that you cannot do the sampling.

Ravi Subramanyan: 00:13:36.204 Last but not least, you have machine data coming in different intervals, different volumes. And so that kind of is — and it's coming in real time too. So you have to be able to collect all this information and be able to present it to be able to do the OEE analysis. So at this point, I'll hand it over to Ryan to talk about our general architecture. Ryan, take it away.

Increasing OEE in the Enterprise

Ryan Dussiaume: 00:14:00.046 Thanks, Ravi. So while our demo is continuing to provision here in the background, we were thinking it would be a good idea to just show you sort of what the demo architecture that you saw earlier from Joe looks like when it's kind of scaled up to do OEE, to increase OEE in the entire enterprise. And so typically what we see is with, let's say, a large manufacturing site that's implementing this kind of architecture, we'll have a FactoryStack of software that's on-site, that's connecting to those on-site assets, such as PLCs, gateways, and sensors. And a key part of being able to increase OEE is getting visibility into that real-time data from those devices. The way that this architecture is designed to do this is through what's called a Reliable Real-time Data Backbone that's essentially enabled typically by an MQTT broker, in this case, HiveMQ. And a little bit later, I'll dig into some details on what the architecture of HiveMQ looks like and some of the details of this real-time data backbone.

Ryan Dussiaume: 00:15:23.019 But essentially, what it is, is all about being able to reliably deliver that data to the cloud in real time so that the FactoryStack in the cloud can leverage this data as well as the applications and cloud services can leverage this data to provide the information to the people that need it in order to take action to increase OEE. And so there's really three pillars to this solution that we're going to be going through now. And this is really about simplifying that provisioning of that FactoryStack. And for that, we're going to go to James from 4IR. And then I'm going to talk about how the HiveMQ real-time data backbone can power this FactoryStack. And then we're going to go over to Remus from Riveron. And he's going to talk about how they help with the integration and also at the backend in terms of leveraging this data in order to take action. So over to you, James.

James Burnand: 00:16:29.219 Thanks, Ryan. So do a little quick introduction about 4IR's FactoryStack and PharmaStack, if you've heard of our products before. They're fundamentally platforms-as-a-service that we deliver as a managed service. And really, the purpose of doing this and why these exist is to make it easier for manufacturers or industrial users to be able to get the advantages that the cloud can offer and make those functions and features available to them. As we're talking today, we're talking a whole lot about OEE, but fundamentally, this architecture is pretty widely used and pretty widely supported for a variety of different data applications. It just happens that OEE is usually the lowest-hanging fruit, or even just performance management in general is the lowest-hanging fruit because it's the thing that can show immediate value and is often very fast to deploy. But we've seen this architecture and this sort of data pipeline that Ryan's going to talk about be used for a variety of different use cases.

James Burnand: 00:17:28.112 So really, if you look at our diagram on the right here, you'll see FactoryStack and PharmaStack. And we labeled them very similarly. They look kind of the same. And fundamentally, they do the same thing, with the differences between the two being that PharmaStack is really designed to be operated in a pharmaceutical-regulated environment. So there's some data integrity considerations, some retention considerations. And when it comes to some of the documentation that we provide as a part of our upgrade and management process, it's a little more verbose and involved for the pharma version, but fundamentally they do the same things. And that same thing is we are able to deploy a set of applications, services, and operationally run and manage these systems for users, both in their cloud, in our cloud, as well as on-prem using hybrid technologies.

James Burnand: 00:18:17.683 So as you can see in the diagram, on the very bottom layer, we show Azure and AWS in our tenant and their tenant on-premises. And then that next layer up, which is a very consistent way of deploying these sorts of solutions. And that really represents a combination of our management plane that we use for deployment, management, monitoring of all the different systems, as well as the bespoke parts of the application that are specific to any one of our subscribers. It doesn't really show it here, but every one of our subscribers gets their own VPC, their own VNet, and their own copy of every one of the services that they're going to be using so that we can create as much isolation as possible knowing that you're still using the cloud, which is technically someone else's computer. The other part of this is that a lot of our customers in the bigger enterprises — they already have private lease lines or express routes or private networking built. So we're able to take advantage of some of those previous investments to create a public cloud offered and managed service that resides and runs across private or at least network communications.

James Burnand: 00:19:22.286 So fundamentally, though, we talk about the applications that we support on the left, the Ignition, HiveMQ database. There are others that we can support natively as well. But really, what we can do is we can run any application inside of FactoryStack or PharmaStack. Those are the ones that we're able to do in a super duper automated way. So as you can see, Joe pressing a couple of buttons allows for us to provision, but also update, manage, and monitor those applications as a part of that same deployment. So the technology that we use is Infrastructure as Code, which may or may not be familiar to everybody on this presentation. But that, plus the combination of technologies that we've chosen, things like Orchestration, allow for us to be able to run and operate these at scale. On the left, I highlight, real quick, our partnerships. So obviously working with HiveMQ, they are the de facto from our perspective. Our enterprise broker, certainly when operating at the kinds of scales that you need to when you're running across multiple facilities with thousands of different endpoints and millions of different data point connectors, really the scalability, interoperability, and the ability for us to include it in our platform and scale it in our platform, it makes perfect sense for a lot of use cases.

James Burnand: 00:20:36.416 We work with Opto 22 from an edge perspective. They've got some really awesome hardware for deployment and applications. And that's part of this demo as well, providing kind of that last mile to some data. And then we are the solution partner for Inductive Automation, which is the application that Remus is going to demo is built in Ignition, which is Inductive Automation software. We're their solution partner for managed infrastructure. And really, they're a leader in the SCADA and I'll call it a plant floor operating system is probably another way you could describe it. But fundamentally, it's not just SCADA. It allows you to build all kinds of different MES and other business applications and plant floor applications and take advantage of that real-time data. I will highlight as well that we've got Microsoft and AWS shown here. Those are the two cloud platforms that we currently support. We are partners with both. And on the AWS side, our applications have passed the foundational technical review from AWS. So they are considered qualified software.

James Burnand: 00:21:38.080 And then our delivery partners, so Riveron and Grantek, are shown on here. There are our primary delivery partners for building entire solution sets for end customers. So being able to not just stand up and build the infrastructure, but go all the way to delivering that end value that the customers need. On the bottom of the slide on the right, you'll see IT/OT consulting and industrial cloud assessments. Really, this is a part of our services that we provide for folks that aren't sure exactly what to do, where to start, how to get started, what is the lowest-hanging fruit, or how can I weave something into my architecture or my plans that makes sense for the future? And that's where we can come in and help with architecting discussions or doing some technical debt identification to figure out where investments could be made to see the biggest bang for the buck. And so for that, I will pass it back to you, Ryan, if you want to keep on your track.

Ryan Dussiaume: 00:22:33.756 All right. Thanks a bunch, James. And so here we wanted to dig in a little bit more on what you would deploy from a HiveMQ perspective as part of 4IR's FactoryStack, which just makes it, let's say, a lot more seamless and part of a larger ecosystem of tooling that can be used for addressing your use cases in manufacturing Industry 4.0. And so the way this looks typically is you'll have essentially a HiveMQ cluster for reliability. So multiple HiveMQ Brokers deployed in a cluster on-site. And typically, this is a smaller deployment because it's really meant to handle the real-time data from the devices within a single site. Then using HiveMQ's Bridge Extension, you can create an MQTT connection. And we advise using MQTT in this regard for going from on-site to cloud because of the reliability that it provides. Oh, sorry. I think we switched slides.

Ravi Subramanyan: 00:23:42.575 Sorry, sorry.

Ryan Dussiaume: 00:23:43.436 Oh, sorry. Thanks. And so, yeah, we advise using MQTT for this bridge because of the reliability that it provides and the dynamic clustering — or sorry, dynamic queuing that it can do to address things like potentially unreliable networking connections or networking drops that can happen between on-site and cloud. And so this is all about reliable data transfer between those two areas. Then we have the HiveMQ cluster in the cloud, and this is typically a larger cluster. HiveMQ can scale to millions of connections and millions of messages per second. And so it can handle the scale where you might have one site, or if you have 20 or 80, it scales to handle that real-time data and do it reliably as well. The next piece here is HiveMQ's support for streaming platforms. So getting this data into the different clouds through their respective streaming platforms. So in Google Cloud, this is Google Pub/Sub. In AWS, this is Kinesis. In Azure, this is Event Hubs. And then typically what we see when customers want to take a little more of a cloud-agnostic approach, we see them deploying and using Kafka.

Ryan Dussiaume: 00:25:11.340 And HiveMQ works really well with these streaming platforms and industry  in particular because of the way that these applications are more about high throughput. And so they're pushing complexity outwards towards devices and applications so that they can have maximum throughput and that they don't do complex routing of messages and they ensure that they have other integrations to many other kinds of data platforms that are within the respective clouds. HiveMQ works really well with that because HiveMQ is all about taking complexity away from clients. And so we work really well with that. And we work in the IoT space, whereas they work in the application space. And so this is a really good pairing, let's say, for that purpose. Go to the next slide. Thanks a lot. And so let's zoom into one of those clusters. And just taking a look at how this looks in detail for things like redundancy and persistence and performance and protection. So a HiveMQ cluster is a masterless cluster. The nodes work together independently of any kind of separate piece that you need to manage. They replicate data throughout in real time so that when you are expecting a Quality of Service (QoS) confirmation that the message has been delivered, with HiveMQ, you can ensure that it will be because of this data replication, meaning if there's a single point of failure in the cluster, the cluster will just continue working.

Ryan Dussiaume: 00:26:55.074 Persistence as well is a major aspect of the HiveMQ platform for reliability because since we persist the data to storage that's in transit, this means that we can handle much longer outages. So when the data is being queued, when there's an outage, this can happen for much larger amounts of data. Also, supporting that redundancy at scale, like we said, scaling up to, let's say, 80 sites connecting into that central cloud broker, this is a necessary piece. Protection, doing things like connection limiting and cluster overload protection, throttling, protecting against connection storms when there are network outages. This is a big part of what makes that backbone super reliable. It's not going to fail in those kinds of failure situations. And performance, being able to do this in this kind of a clustered environment and perform at scale, there are special aspects of the platform that enable this to happen. And that's, in particular, the fact that data is referenced with a consistent hashing between cluster nodes.

Ryan Dussiaume: 00:28:08.911 So as your cluster scales and grows, you continue to have the same reaction time on every single request that comes into the cluster. So at the end of the day, this is all the aspects of HiveMQ that go back to supporting that reliable real-time data backbone or Industry 4.0. Passing it on to Remus now. He's going to talk about how to leverage the software stack for OEE and for other purposes and how they help with that.

Remus Pop: 00:28:43.050 Awesome. Thank you, Ryan. Thanks for going through that, and James, for going through your piece as well. So like you heard earlier on, Riveron is a national business advisory firm. We do business across multiple different business sectors. My group, specifically the Intelligent Manufacturing Solutions Group, focuses on Industry 4.0, performance improvement, and what we call manufacturing transformation. So a lot of what I'll cover talks about the entire circle of what we do for OEE, right? We're going to focus a lot on the technology here in this demo and what you're seeing from 4IR and Riveron and HiveMQ. But a lot of what it takes to increase that OEE number comes from the people and process side. So we understand, as a transformation team, that beyond just the technology piece, we have to understand that the technology is just a part of it, right? If we deploy this OEE dashboard that you're going to see shortly and it shows you all of the data in the world, it doesn't do anything on its own. So we have to understand, as a company and as an organization, that we have to take that data and create action plans from it. So the data is just there. Gives us insight into what we can focus our attention on as an organization.

Remus Pop: 00:29:51.472 So Riveron as a whole, we have about 12 offices across the country. We have a little over 700 advisory experts, again, across multiple different segments. My group specifically focuses on Industry 4.0. We come from industry. We have a long background in manufacturing and connecting equipment. And if you flip to the next slide, I'll kind of talk a little bit about our process into going into this type of environment. So one of the things we like to look at is understanding that it's a lot to take on a transformation project right off the bat, right? When we talk to our customers, they're looking to just begin on this journey. They're not fully bought in, but they have some desire to improve their efficiency, maybe understand a little bit more about what's happening in their factories. And so one of the things we like to look at is attempting a pilot. And, right, with partners like HiveMQ and 4IR, the path to getting to a successful pilot becomes a lot slimmer because I don't have to worry about the entire infrastructure being available, right? So a lot of the things that hold us up in a factory setting are having to provision VMs and set up a database and set up Ignition and get connection to all these different things.

Remus Pop: 00:30:56.451 And so one of the things that I think is really impressive with what 4IR and HiveMQ can provide is being able to lower the cost of entry for completing a pilot. So what we want to do is we want to look at an area in our facility that needs attention. Maybe an area that is struggling to get parts out the door. Maybe their quality numbers are at the bottom of the end. What we want to do is focus our attention on something like that. Because one of the things that we've noticed through the years is that if we look at doing a pilot on a piece of equipment that maybe only runs a couple days a week, it's off in the corner. It's not really super important, so if we mess up, it's not a big deal. Well, that's great. But also, if we're really successful, it's also not a big deal because that machine is only running a day or two a week. But what we want to focus on with a pilot is looking at a high-value asset that is struggling. Because if we get 1 to 2 percent on a machine that's producing 50,000 parts a day, that 1 to 2 percent becomes a pretty big number, right? So that's what we want to focus on when we're looking at where we start this pilot. How do we take OEE data and impact our production numbers?

Remus Pop: 00:31:59.362 But then, of course, the big part is the connection. Now that we have this targeted line, we want to look at: What do we have to do to get the data out of that machine, right? We take a team and we look at what type of controller we have. Was it built in the era of ethernet? A lot of times, manufacturing facilities that we come across, there's still World War II era equipment in there. It doesn't mean we can't get data. It just means we have to get a little bit more creative on the way. And we highlighted the Opto 22 partnership. That's where that really starts to come in and be super powerful because we can take an IoT-enabled device with Ignition running onboard, connect it to a machine that was built in the '50s via some sensoring, and get some actionable data. We're not going to get the full breadth of what we might expect from a brand new Rockwell or brand new Siemens, but it'll at least allow us to gain some insight into how that piece of equipment is running to understand its throughput, its utilization, its availability. We can start to look at some of those metrics that make up OEE that will let us start to build this story of what we want to do to improve our manufacturing operations.

Remus Pop: 00:33:01.452 And then now that we have this piece of equipment connected via direct connection or with enabling it with an Opto 22, now we look at the digital integration piece of it, right? This is where we start to look at the 4IR and the HiveMQ piece. We have this platform in the plant. We have machines connected. Now what do we do with all this data? Well, with what you're going to see here shortly — is we're going to move that data to HiveMQ, pull it into Ignition, and build some visualization on it. That's really ultimately what we want to get to, right? We've got to get the data backbone set. Like Ryan mentioned, we got to do that high reliable data backbone, get it into HiveMQ so that we can start to then rapidly deploy a visualization tool on top of it. And we use Ignition on both ends of it, right? We use Ignition on the plant floor to collect the data. And we use Ignition on the cloud side to visualize it. And what you're going to see here is an application that was built in Ignition, looking at tools that we have available to us with the perspective module and some of the connections into HiveMQ via MQTT engine and MQTT transmission, some of the SiriusLink modules. And we're going to be able to show you what Joe just stood up about — what time is it? About 30 minutes ago, we're going to be able to launch an Ignition application and actually have some real data getting fed into that.

Remus Pop: 00:34:11.197 And then, of course, once we do this successful pilot, we've built this data integration pipeline, we've built this backbone of foundation, we've created a cool dashboard, now we have success, right? Now we can take a look at this data and start to put into practice of understanding what to do with this. And the way we typically evaluate that is from a persona perspective, right? Again, we go back to the people-process piece. So the people part of it is everybody within an organization has a different need for data, right? Your COO might require different data than a maintenance tech, than a plant manager, than a process tech, than an operator. We want to make sure that we understand that the goal of Industry 4.0 (I've said this multiple times) and I think other people have echoed it quite regularly — is that the goal of Industry 4.0 is to deliver information to those who can act upon it the fastest. Whether that's an operator, whether that's a CFO, a CIO, a CTO, CEO, right? The whole goal of what we do is to build this data integration and this digital integration into a lot of the different platforms that you're going to see here so that we can understand what data can benefit what person and what we can present to them so that they can do their job better.

Remus Pop: 00:35:17.344 And then again, what Riveron can help with is that end-to-end transformational piece of, right, using this dashboard in your standup meeting every morning to understand what we need to focus on from the previous day's shift. Teaching your maintenance team how to pull a downtime report so that they can see on Machine 7 — or, "Line 7, Machine 5 has had 47 faults in the last 24 hours for this one thing. Maybe we should take a look at that switch or that station or that valve or that pump. Something is going on, on that station that we need to focus our attention on." And what this does is kind of give us those tools to enable that sort of intelligence into what's happening. So now I am going to jump into the demo —

Demo of Zero to OEE (OEE Application Demo)

Ravi Subramanyan: 00:36:01.396 All right, I'm going to stop sharing now. 

Remus Pop: 00:36:03.095  — and share my screen. [crosstalk]. Give me one second. Hopefully I can do this pretty smooth here.

Joseph Dolivo: 00:36:09.586 Yeah. Remus, I'll actually start out by showing kind of the resources up and then I'll show the admin console as well.

Remus Pop: 00:36:15.746 Absolutely. While you're doing that, I'm going to get the app open and log in.

Joseph Dolivo: 00:36:18.780 Sounds good. So this is pretty cool. This is actually showing the results of the — we basically have three linked pipelines that deploy the entire infrastructure. And you can see, these all ran 11 minutes plus 7 minutes plus 5 minutes. So we're at about 23, 24 minutes with the seconds in there. So we provisioned all the stuff that I had shown you from scratch. And so we were basically using this zero-to-OEE subdomain. So if I go over here, you'll see this is actually the broker itself. And we dynamically generate credentials so that they are secure and unique for every individual instance. I'm going to show you this really neat tool called basically the management console, the Control Center that's provided by HiveMQ. You can see this has been running for just a couple of minutes. And this gives me some really good insight into the status of the cluster. So I can see how many connections there are. I can look at things like publish and subscribe rate. I can look at the number of devices that are subscribed. This is basically giving me connections by node because I am running in a two-node cluster right now. I can obviously have more of those. I get some statistics per cluster down here for each of the different nodes.

Joseph Dolivo: 00:37:27.595 And I also can actually get a snapshot of all the clients that are connected. So I can come in here, refresh, and then I can see basically what's up in here. So it's pretty nice to be able to have this tool where I can take a look at all of this. And it just makes it really easy for standing up and for troubleshooting. And it's, like I said, a nice visualization tool. So that's what I wanted to show, Remus, so feel free and go ahead and go over to the Ignition application.

Remus Pop: 00:37:57.337 All right. So what we're going to see here is that instance of Ignition and that gateway backup that Joe just provisioned. So keep in mind — this is an Opto 22 running on James's desk. So it's not real manufacturing data. I don't think we could have had anybody sign up to allow us to connect to their production system in real-time for a webinar demo, but you'll get the idea of what typically we show. So typically, when we land on the application, we look at a high-level view of — our Birmingham office is what we'll call it. So my office here out of the Detroit area is based in Birmingham. So we're looking at Birmingham. We're looking at, in this case — availability is important to me. This number could be OEE. It could be performance. It could be quality. It could be number of pieces made. The idea here is that we want to show the KPI that's important to the specific customer that we're focused on. So in my case, I want to look at availability. And within my Birmingham facility, I have Area 1. I could have Area 2, 3, 4. The numbers continue to go on.

Remus Pop: 00:38:52.957 Now I have two departments. I have Department 1 and Department 2. So I'm going to look at Department 2 because that's running at about 100% availability in my test line. So what we're going to do here — you'll see that over the last few seconds, we've had this PLC get connected and it's streaming a ton of data. So we're getting all kinds of values off of this machine. We want to look at different types of characteristics and to see what is happening on that machine. In this case, we've been [inaudible] since it came up, which is about a few minutes ago. We're in a pretty unique state here. So we want to start to look at different values and different characteristics of what happens. Again, this data is coming from a sample PLC, but the idea here is that we look at a shift target. We have an idea of how many we want to make this shift. We have a pretty good idea of where we are right now. So far, since Joe turned on that Ignition gateway, we've made 3,400 pieces. We're moving pretty quick. Our delta is pretty off, based off of the output, right, in a ton of scrap. And then again, we look at our performance and availability or some pretty crazy numbers there based off of what we've seen. But the idea here is that we have multiple different views to create data for operators that might benefit from it, or maintenance techs, or CFO, COOs. Again, it's all kind of built around the idea that we want to deliver the information to those who can act upon it the fastest.

Remus Pop: 00:40:10.123 And so here you see a little bit of different type of information that's getting presented. We're looking at a status line, Machine 1, 2, 3, 4. We have what's pretty popular in our space, which is an Andon board, which is just showing you shift over shift, hour over hour, what is in production. We have the ability to build into different types of reporting tools like a line status machine. We're going to look at current shifts so far and just get an understanding of what's happening in a ribbon format, just different ways to display data. Again, the idea here is that we want to create an environment and a template that allows our customers to move and create their own instance of what they value as an OEE platform. So again, as anything else, we want to know how many parts did we make over time? Obviously, this instance just spun up. So we got to do some pretty interesting things, Test Line 2 and Test Line 1. All the different things we can do here to look at data. Downtime instances. We want to make sure that we have different tools available to us to look at downtime. And again, this is all still getting provisioned in the back end. So we want to make sure that, over time, this will get more rich with data as we start to look at a lot of the different things that we got.

Remus Pop: 00:41:21.608 But the idea here is that we have a tool that's able to get deployed rapidly with all of the things that you saw earlier to look at different capabilities and different metrics. That's the idea here — that with everything that we had done, 30 minutes ago this didn't exist. This gateway backup didn't exist. The infrastructure connected to HiveMQ and the Opto 22 didn't exist. None of these tools existed for us 30 minutes ago. And we were able to deploy this entire instance that was able to connect to a HiveMQ cluster, pull in that data, start populating screens in almost no time at all. Now, I know we got about two minutes left before we jump into the Q&A session. So in preparation for this, we did set up an instance that might have a little bit richer data. So I'm going to pull that over so that you can see what those dashboards would look like in an instance that may have been running for a little bit longer, and we get a little bit more robust data sources. So now we can see that we've seen up and down. We get some little bit more accurate numbers as this has been running a little bit more efficiently. Different Andon characteristics, different report data we can pull back, downtime instances, different things like that. So we see here that if I look at Test Line 1, I see Machine 1 had about 12 occurrences of downtime. We lost 3.6 minutes of production.

Remus Pop: 00:42:44.856 So now if I look specifically at machine one, I can see what those were, right? So now I see I had six instances of a breakdown, six instances of automaterial. Here's the different timestamps that went along with those. We get the unique case in some instances where our operators and our teams want to be able to manually enter data. Maybe in one of those cases, like we talked about earlier, it's legacy equipment. And all I'm getting is on/off signals. So in this case, I want to go back and say this actually wasn't a specific reason, but I want to change it to — maybe it was too many widgets, or we didn't have enough widgets, or maybe someone ran into the machine with a high low. So we have the ability to then, again with the ignition platform and the database structure that we've set up, go in and edit some of those reasons based off of things that might have happened on the shop floor that the machine itself couldn't automatically detect. And with that, I think that comes up to the end of the demo. And we'll get into the Q&A portion.

Ravi Subramanyan: 00:43:38.936 Yeah, yeah, great. Thank you. Thank you for running through that wonderful demo. Let me take back the screen here and share my screen here. Share.

Jayashree Hegde: 00:43:51.305 Yeah. Everyone, attendees, if you like the demo, please give us a thumbs up. And while Ravi is setting up the Q&A slide, I would like to launch the poll. I request all the attendees to cast your votes. Thank you.

Q&A

Ravi Subramanyan: 00:44:08.621 All right. So should we just jump into the Q&A at this point or should we give them a little time to answer the questions?

Jayashree Hegde: 00:44:14.528 No, I'll run the polls till the end of the session so we can go ahead.

Ravi Subramanyan: 00:44:19.719 Okay. Okay. So we have a few questions that have come up, so let's jump right into them. So I think this one is relevant to you, Remus. You just mentioned this. So once you have the POC-POV completed and the company is seeing value, right, recognizing the value from the process, we now have an IT problem, which is governance. What is the typical process of bringing the POC into an ongoing business process? Maybe Remus, you can take this question.

Remus Pop: 00:44:47.810 That's a great question. Yeah, absolutely. So that's a challenge we see in every one of our customers, right? We do a pilot. It blows everybody's socks off. And now we want to start deploying. And the operations team always wants to move really quickly, "Oh man, this is great. I want to put this in every one of my plants. How do I get it into my facility in Puerto Rico, in Mexico, in China? I want to get this everywhere." And then IT comes in and says, "Whoa, whoa, whoa, wait a minute. You're going to start deploying all this new technology all over the place? What's your change management process? What's your governance model?" Right? There are tools that we have within our firm that outline what is the best practice for this. But ultimately, it's to the customer to work internally to understand what matters to them, what change process or change management process is going to help their organization manage this as it continues to roll out. There's not only change management and all those things, but there's also training. How do you train your operators, your maintenance techs, your facility leaders to use this new dashboard?

Remus Pop: 00:45:42.165 So there's a ton of different things. And there's a framework well established that we've used with a number of our customers, but ultimately, at the end of the day, it has to be integrated into your company's business process. And yes, it is ultimately ongoing forever, right? This is not a project that we're going to do one time and then we're going to call, "Okay, cool. We did digitally transform." When you decide to take this initiative on, you are fundamentally changing the way you operate as a business. And as part of that, you have to understand that there are different tools that are available, different pieces of technology that you integrate. There's processes that are going to change. All of these different things are going to come into play as you look at taking this beyond POC into real digital transformation.

Ravi Subramanyan: 00:46:26.920 Awesome. Great answer. Thank you, Remus. Anybody else wants to chime in? I think, Remus, you've summarized it well. So let's jump to the next question. So this is specifically about MQTT. So maybe Ryan, if you want to take this. How ubiquitous is MQTT as a protocol for IoT and Industry 4.0, 5.0? Or is there fragmentation there?

Ryan Dussiaume: 00:46:52.084 Yeah. I mean, obviously as an MQTT broker and working in a company that builds an MQTT broker, we may have a bit of a skewed view on this, right, because we are seeing a lot of questions and interests and opportunities in using MQTT for this purpose. And it's just catching fire, if I could be frank. It's becoming quite ubiquitous from what I can see in many different verticals, not only manufacturing, but also things like oil and gas, for example. And so there are so many aspects of the MQTT protocol that are beneficial for this purpose, right? It's very lightweight. It's very easy to scale up new use cases on top of an existing platform just because of the way that the MQTT protocol is designed to work in an IoT space, meaning that it just handles that variability that comes with a number of topics, the number of connections, the types of devices, the types of payloads. And just handling that uncertainty of the networking environment in those IoT environments is a big aspect of the protocol. And so it's just really addressing this need. And we see it kind of growing extremely quickly in this space. But happy if someone else wants to comment on that as well.

Ravi Subramanyan: 00:48:29.645 Yeah, I can jump in, right? So one of the things that we see, yes, we are an MQTT broker company, so we are a little kind of biased here. But MQTT is rapidly becoming the de facto edge-to-enterprise or edge-to-cloud protocol because of the advantages it brings, be it the ability to work in constrained environments, limited bandwidth, limited connectivity. The public-subscribed event-based or exception-based ability to send the data, which is very efficient based on the fact that it's based on TCP, which is pretty robust. So it is gaining a lot of traction in terms of edge-to-cloud or edge-to-connectivity. There's still some fragmentation on the south side. But I think, as Remus and others mentioned, there are solutions that can try to consolidate that, right? There was a follow-up question about, "How do I get more training about MQTT and about HiveMQ?" I think the best way is HiveMQ kind of believes in MQTT, right? So we put out a lot of content. So there is a lot of content available. I think we have — let me see. Yes. So here are some of the resources that you can immediately get access to. There's a lot more resources on our website that talk about MQTT essentials.

Ravi Subramanyan: 00:49:42.387 We didn't touch upon Sparkplug, but Sparkplug is another data framework on top of MQTT that provides additional feature functionality for manufacturing that you can also get information on. I think the best way is to download our software. It's free. You can start downloading it and then start playing with it. Of course, I mean, as you kind of specifically want to work on use cases, you can work with the team, right, because it's important to figure out what your goals are and how we can work through. So certainly, please reach out to us and we are happy to help further. All right, so let me go back to the next question. So maybe this one can be taken by the 4IR team.

James Burnand: 00:50:21.388 Hey, Ravi, can I just give a bit of perspective on the last question?

Ravi Subramanyan: 00:50:24.230 Of course. Of course. I'm sorry.

James Burnand: 00:50:24.410 I think I haven't talked enough in this webinar.

Ravi Subramanyan: 00:50:27.406 Yeah, yeah, yeah, yeah. For sure, for sure, for sure.

James Burnand: 00:50:30.428 I would just say that we spend a lot of energy and time with customers, thinking and looking through different architecture options and different potential ways to handle things and to scale things. And really when we start looking at some of the benefits that you get from an MQTT-based architecture, for example, being able to have read-only connections and being able to enable QOS for certain use cases and being able to have highly distributed — whether I've got facilities that are really new and modern and well-built versus ones that are more challenging from a communication/network reliability perspective or from a modernization perspective, MQTT is kind of an equalizer in that space in that it allows for data models to be created. It allows for low bandwidth connections, reliability. They're all options that you're able to choose and build. So it's not the only technology out there, but almost everywhere we go, it's winning the race.

Ravi Subramanyan: 00:51:34.844 Great. Thank you, Joe. So maybe you or James can take the next question, which is basically, "How much does this business — how much of this is driven by product companies, meaning product companies running it themselves, versus working with consulting companies or solution companies, like infrastructure companies such as you, right? Where does that really come to fruition or benefit?"

James Burnand: 00:51:58.641 So I would say that this entire space is being driven by digital transformation. So however that happens to kind of rear its head in an organization. It's usually an executive-level sort of an initiative that doesn't always have Industry 4.0 as the starting point. A lot of times there's digitalization around HR and around how our supply chain is going to be connected. And then all of a sudden it's, "Well, we need manufacturing data to be able to complete this picture." So I think, if I kind of look at it from an OEE perspective, it's also usually connected, at least these days, to a digitalization and information centralization and being able to apply digital technologies to manufacturing. I think that's often the reason why it gets started. And there's different companies out there that have — there's different philosophies for how to gain the value of OEE. And I think as Remus explained, it's not a piece of software that you buy and it just magically makes your world better. It's a culture shift. It's an adoption of technology and a sponsorship from a leadership perspective of a methodology. And that's a concept that I unfortunately think sometimes gets short-changed in the way people talk about it.

Remus Pop: 00:53:20.472 Yeah, I think I can say as well too, both from the consulting side and not long ago being the customer on the manufacturing side. If you look at whether it's a product company or a consulting company, I think it depends on your organization, right? We have a lot of customers that come to us for advice and strategy around how they can do it themselves. So what we would do in that case is recommend product companies like 4IR, like HiveMQ, based on the business use cases that they have. And then we have companies that come to us for an end-to-end solution. They want us to integrate, build the strategy, find the partners, integrate the partners, do the whole thing. So I think really what it comes down to is the organization, the customer, and what the goals of the customer are. Do they have a team internally that knows how to do cloud provisioning, like Joe and James kind of showed off a minute ago? In manufacturing, we typically see that they don't. So there's opportunities, I think, on both sides, both from a product side and a consulting side. It just really depends on the customer and what they're trying to accomplish.

Ravi Subramanyan: 00:54:20.089 Okay. Great. Thank you. And I know there's a lot of questions coming, and thank you so much some of you for answering the questions. So here is one. I know we have like four minutes left, but I think we can tackle this, right? "I've heard of different solutions for creating OEE monitoring such as StatSoft, Ignition. I think we talked about Kepware. Is any of the — is there any more of them that you can recommend?" I think this is kind of like what we are proposing in terms of architecture. There's multiple ways to do it, but you want to go with something that's proven to where we can bring together different solutions to help you achieve this, right? So that would be my response. Remus, being in the consulting world, would you like to chime in?

Remus Pop: 00:55:07.558 Yeah. So I think I'll echo what Arlen typically says about MQTT. And I'll say that, "One of the cool things about Industry 4.0 is that there's a lot of companies interested in this space and creating really awesome solutions. The bad thing about Industry 4.0 is that there are a lot of companies in this space creating really awesome solutions." So trying to navigate through what is the best one is really difficult right now. So what I would suggest is to create a set of requirements that matter to you as a company and then evaluate all of the different solutions, whether it's StatSoft, ignition, ThingWorx, Litmus, Tulip, just to name a few, right? So create the requirements for you as an organization and what matters to you, and then use that to evaluate the different solutions that are available.

James Burnand: 00:55:49.406 I would add to that, though, Remus, that I think there's a difference between evaluating for a particular application versus evaluating a platform. Because everyone you mentioned there is a platform. There are certain platforms that are better at certain things. And so when you think about a digitalization platform, if you have a high level of sophistication and automation, there's probably some software that's going to be better for you than if you're a lower level of automation and sophistication and you're looking for more paper on glass sort of [crosstalk].

Remus Pop: 00:56:16.341 Yeah. Exactly. Yeah. Exactly.

James Burnand: 00:56:17.522 Yeah, we have this conversation often — is that you can't really compare. I'll use Ignition as an example. There are hundreds of system integrators who have built different types of OEE applications. Plus there's some built-in modules for Ignition for OEE as well. So comparing one of those versus a bespoke OEE application that you purchase from a vendor doesn't really take into account what the value of a platform will be, and being able to employ different types of applications on those platforms that meet different business needs for you.

Ravi Subramanyan: 00:56:51.105 Yeah. And I know we discussed it offline too, but it's all about the people, process, and technology coming together, right? I mean, technology is there, right? We can provide all the information, but if you don't take action on it, you're not going to be able to improve OEE, right? And then working with a solution provider such as us, we'll make sure that we not only help you with the technology, but we'll also help you with the people and process things so you can actually get a benefit out of — or change behavior, right, to be able to achieve your goals.

Remus Pop: 00:57:23.934 Absolutely.

Closing Words

Ravi Subramanyan: 00:57:24.557 All right, I think we have just a couple of minutes. I'm not sure if we can address questions here. So Jayashree, at this point, I think we can end this. And obviously, there's been a lot of other questions, so we can certainly make sure that we respond to all your questions. I want to hand it back to you, Jayashree, for any final thoughts.

Jayashree Hegde: 00:57:41.782 Thank you. Thank you so much. Thank you so much. This was an awesome presentation. Thanks, James, Joe, Remus, Ryan, and Ravi for this awesome, awesome presentation and the demo. And thank you to all the attendees for tuning in. We hope you all enjoyed it. Like we already shared in the beginning of the session, we will send a follow-up email with the recording as well as the slide presentation. And for the follow-up questions, you can always reach out to us. We have included the contact details of all the speakers in the last slide. And feel free to submit the contact us form. So thanks again for tuning in. Take care. See you all next time.

Ravi Subramanyan: 00:58:24.676 Yeah. I think there was a question on how do we get started? I think you can contact any one of us and then we can bring in the other team. So I think that should be easy enough. So any of us should be able to get you started.

Jayashree Hegde: 00:58:35.575 Yep.

Remus Pop: 00:58:37.929 Thank you, everybody.

Jayashree Hegde: 00:58:38.737 Thanks, everyone.

S0: 00:58:39.210 Thanks, everybody.

Ravi Subramanyan: 00:58:39.679 Bye-bye.

Jayashree Hegde: 00:58:40.521 Bye-bye.

Remus Pop: 00:58:41.800 Bye-bye.